Abstract
The central Southern Alps of New Zealand/Ka Tiritiri o te Moana is a region of active geomorphological change. It experiences between 5-10 m of precipitation annually in the form of snow and rain that generates both glaciers and potentially torrential river flows. Its position adjacent the Alpine fault exposes it to regular high-magnitude earthquakes, numerous small-scale earthquakes and steady annual ground deformation. Prolonged glacial recession in this region can lead to debuttressing as the footwall support of the mountainsides is removed. These drivers of change are profoundly reshaping the Alpine landscape and can contribute to various types of slope instability and slope failure. This thesis leveraged recent progress in Aerial and Satellite Photogrammetric Mapping (APM/SPM) to detect landscape change and inventory slope failures in the Aoraki/Mt Cook National Park over the 2008-2017 period.
This thesis first characterised the performance of SPM with Pléiades 1A satellite imagery by testing the accuracy of topographic mapping with a tri-stereo image of the Franz Josef area. Three georeferencing scenarios were tested that would affect the accuracy of Digital Surface Models (DSMs). 1) A fully automated processing using only vendor-supplied georeferencing in the form of Rational Polynomial Coefficients (RPCs) that resulted in an average 4.4 m planimetric offset, and a tilt causing vertical offsets reaching up to 20 m; 2) processing using additional Ground Control Points (GCPs) collected in the field and sampled from another Southern Alps photogrammetric block that resulted in 0.48 m RMSE planimetric accuracy but also a vertical tilt and up to 1.5 m of resulting offset; and 3) processing via incorporation into a broader photogrammetric model that produced no systematic vertical offsets as well as sub-meter (0.48 m RMSE) planimetric accuracy suitable for change detection.
APM/SPM using the final georeferencing scenario was then applied to process another Pléiades stereo-pair of the Aoraki/Mt Cook area captured in 2017. The resulting DSM and orthoimage were compared to a those derived from an aerial survey captured in 2008. The 2008-2017 DEM of Difference (DoD) was used to detect, map and characterise new slope failures, and quantify volume change. 557 previously unknown slope failures were found and mapped over the 8.5-year period, covering 11% of the study area, while only 8 were previously documented during 2008-2017. From the analysis of the DoD, this thesis estimates a total a volume associated with slope failure ranging between 0.320 km^3 (estimated directly from the DoD), and 6.555 km^3 (with additional geological interpretation). It also suggests that the majority of landslide activity is associated with glacier downwasting and snowfield melt. Three newly-discovered large (> 10^5 m^3) rock avalanches are presented and characterized as case studies for the discovery and interpretation of landslides via repeat photogrammetric mapping. Finally, morphometric attributes associated with the landslides, such as slope, elevation, and proximity to glaciers were derived from the photogrammetric products to analyse controlling variables likely to predispose an area to landslide occurrence. A logistic regression model suggested slope angle as the main controlling variable of landslide occurrence, although low overall pseudo-r squared values suggested a relatively marginal model performance.
This research demonstrates the potential of repeated high-resolution topographic mapping to quantify landscape changes in the Aoraki/Mt Cook National Park and improve the understanding of corresponding hazards.